CN106326996A - User load prediction method based on electric quantity information - Google Patents

Abstract

The invention provides a user load prediction method based on electric quantity information. The user load prediction method comprises the steps that step 1: an electricity consumption set Q of a user peak load period is determined; step 2: the time number i of the user peak load period is set as 1; step 3: a use probability typical value matrix PBi-1 is assigned to a use probability typical value matrix PBi; step 4: a load curve sequential value set Pi is constructed; step 5: the use probability typical value matrix PBi is optimized by using a genetic algorithm; step 6: i is set as i+1; and step 7: a prediction user load value is calculated by using a Monte Carlo non-sequential random sampling method according to the optimized use probability typical value matrix PBi and the number of various types of household electrical appliances possessed by users. Compared with the methods in the prior art, the user load prediction method based on the electric quantity information is convenient and rapid so that the prediction accuracy is enhanced, a foundation is laid for reasonable configuration of transformer area power distribution transformer capacity, and the power distribution transformer area economic operation level and the construction efficiency can be enhanced.

Description

A kind of customer charge Forecasting Methodology based on information about power

Technical field

The present invention relates to field of power, be specifically related to a kind of customer charge Forecasting Methodology based on information about power.

Background technology

Along with the fast and stable of economic society develops, novel urbanization is carried out, in particular with agriculture in order with beautiful rural construction Being continuously increased of village's residential households household electrical appliance, China's residential electricity consumption load constantly increases, and resident living power utility characteristic is the most constantly sent out Changing.Accurately grasp the growth pattern of customer charge, particularly peak load period customer charge and by electrical characteristics, be reasonable Carry out power distribution station planning, optimize power distribution station and run the basis controlled.At present actual power distribution station planning construction process to Family power load is typically based on experience and estimates, accuracy is the highest, causes power distribution station just to carry out after increase-volume the most heavily mistake Situation about carrying happens occasionally.Therefore, accurate simulation and prediction customer charge change are badly in need of solving.Power distribution station is for user at present The most accurately know power consumption information, spike period every day, peak period, flat section and paddy section can be gathered by user's intelligent electric meter Electricity, additionally can obtain rated power and the owning amount of user household electrical equipment by statistical method.

Accurately grasp the growth pattern of customer charge, particularly peak load period customer charge and by electrical characteristics, be rationally to carry out Power distribution station is planned, optimizes power distribution station and runs the basis controlled.In actual power distribution station planning construction process, user is used at present Electric load is typically based on experience and estimates, accuracy is the highest, causes power distribution station the most heavily to be transshipped after just carrying out increase-volume Situation happens occasionally.Accordingly, it is desirable to provide a kind of convenient, fast, accurate simulation and the customer charge of prediction customer charge change Forecasting Methodology, the reasonable disposition for platform district distribution transformer capacity lays the foundation, and improves power distribution station economic operation level and construction Efficiency.

Summary of the invention

In order to accurately grasp user power utilization load during the planning construction of power distribution station, the invention provides a kind of based on electricity letter The customer charge Forecasting Methodology of breath.

The technical scheme is that

Described method includes:

Step 1: obtain the kind set DQ of user household electrical equipment, power set PDQ, magnitude-set NDQ, use probability Initial representative value matrix PB⁰, the peak load period moment set T and user peak load period power consumption set Q；

Step 2: setting user's moment sequence number i=1 in the peak load period, 1≤i≤m, described m are the peak load period Moment sum；

Step 3: probability representative value matrix PB will be used^i-1Assignment is to using probability representative value matrix PBⁱIn；Described PB^i-1For t_i-1 The household electrical appliance in moment use probability representative value matrix, PBⁱFor t_iThe household electrical appliance in moment use probability representative value matrix；

Step 4: according to described use probability representative value matrix PBⁱCalculate the load value in each moment, structure in user's peak load period Build load curve sequential value set Pⁱ={ p₁,p₂,...,p_i,...,p_m, p_iFor t_iThe load value of moment household electrical appliance；

Step 5: with using probability representative value matrix PB described in genetic algorithm optimizationⁱ；

Step 6: moment sequence number i=i+1 of peak load period is set；

Step 7: if i≤m, then return step 3；If i is ＞ m, then according to the use probability representative value of the household electrical appliance after optimizing Matrix PBⁱWith the quantity of all kinds household electrical appliance that user has, the non-sequential arbitrary sampling method in Monte Carlo is used to calculate pre- Survey the load value of user.

Preferably, the kind set DQ={dq of household electrical appliance described in described step 1₁,dq₂,...,dq_j,...,dq_n, n is domestic The type sum of electrical equipment, dq_jTitle for jth type household electrical appliance；The power set of described household electrical appliance PDQ={pdq₁,pdq₂,...,pdq_j,...,pdq_n, pdq_jMean power for jth type household electrical appliance；Described household electrical appliance Magnitude-set NDQ={ndq₁,ndq₂,...,ndq_j,...,ndq_n, ndq_jQuantity for jth type household electrical appliance；Described height The moment set T={t of peak load period₁,t₂,...,t_i,...,t_m, t_mIt is the moment value of 24 times processed, 1≤t_m≤24；

The use probability initial representative value matrix of described household electrical appliance

DescribedFor t_iMoment jth type household electrical appliance dq_jUse probability representative value initial value；

Preferably, user power utilization duration set Q={q in described step 1₁,q₂,...,q_k,...,q_d, q_kFor the power consumption of kth day user, D is total natural law that intelligent electric meter gathers described power consumption；

Described step 4 calculates t with the non-sequential arbitrary sampling method in Monte Carlo_iThe load value p of moment household electrical appliance_i；

Preferably, described step 5 uses genetic algorithm optimization t_iThe use probability representative value matrix PB of moment household electrical applianceⁱ, including:

Step 5-1: setting the initial value of iterations h as 0, the maximum of iterations h is maxnum, the quantity of population scale For ZQnum；By t_iThe use probability representative value matrix PB of moment household electrical applianceⁱIn each chromosome using probability to carry out correspondence Coding, produces initial population ZQ⁰；

Step 5-2: described iterations h=h+1 is set；

Step 5-3: by population ZQ^h-1Data assignment to population ZQ^h；ZQ^h-1It is the population of the h-1 time iteration generation, ZQ^h It it is the population of the h time iteration generation；Set population ZQ^hThe initial value of middle chromosome marker position l is 0；

Step 5-4: described chromosome marker position l=l+1 is set；

Step 5-5: the l chromosome is carried out Gray code, obtains t_iThe use probability of moment all types of household electrical appliance；

Step 5-6: according to described t_iThe use probability of moment all types of household electrical appliance, with the non-sequential arbitrary sampling method in Monte Carlo Obtain t_iThe load value p of moment household electrical appliance_i, and update described load curve sequential value set Pⁱ={ p₁,p₂,...,p_i,...,p_m}；

Step 5-7: calculate user's bearing at peak load period household electrical appliance according to the load curve arrangement set P after described renewal Lotus average value P av, $\mathrm{Pav}=\left(\underset{i=1}{\overset{m}{\mathrm{\Σ}}}{p}_i\right)/m;$

Step 5-8: obtain the average load set of user's peak load period according to power consumption set Q PG={pg₁,pg₂,...,pg_k,...,pg_d}；pg_kFor the average load value of kth day user, d is that intelligent electric meter gathers user power utilization Total natural law of amount；

Step 5-9: calculate fitness function value ε of l chromosome,

Step 5-10: if chromosome marker position l ＜ ZQnum, then return step 5-4；If chromosome marker position l >=ZQnum, Then perform step 5-11；

Step 5-11: by described chromosome according to the ascending arrangement of the numerical value of fitness function value ε, choose the 1st to ZQnum Step 5-12 is performed after individual chromosome；

Step 5-12: to population ZQ^hCarry out parents' Shuangzi single-point gene Integral cross computing；

Step 5-13: chromosome is made a variation, and update described population ZQ according to the chromosome after variation^h；

Step 5-14: if iterations h ＜ maxnum, then return step 5-2；If iterations h >=maxnum, then by institute The chromosome stating fitness function value ε optimum carries out Gray code, obtains t_iThe use probability optimization value of moment household electrical appliance；

Preferably, in step 7, the non-sequential arbitrary sampling method in described Monte Carlo calculates t_iThe load of moment household electrical appliance Value p_i, including:

Step 7-1: set the initial value of stochastic sampling number of times a as 0；

Step 7-2: stochastic sampling number of times a=a+1 is set；

To household electrical appliance at t_iThe running status in moment carries out a time stochastic sampling, obtains jth type household electrical appliance dq_jAt t_i The a time stochastic sampling running status coefficient in moment

If the random number of a time stochastic samplingThenIf the random number of a time stochastic sampling ThenFor t_iThe use probability of moment household electrical appliance；

Step 7-3: calculate user at t_iThe load value of a time stochastic sampling in moment

Step 7-4: if stochastic sampling number of times a ＜ RS, then return step 1-2；If stochastic sampling number of times a >=RS, then calculate RS Described load value p under secondary stochastic sampling_i；

DescribedRS is largest sample number of times.

Compared with immediate prior art, the excellent effect of the present invention is:

A kind of based on information about power the customer charge Forecasting Methodology that the present invention provides, make use of peak every day that intelligent electric meter gathers Load period electricity, uses genetic algorithm to be optimized the use probability of all kinds household electrical appliance；Carry out the non-sequence in Monte Carlo Pass through sampling simulation power load distributing.Can convenient, fast simulation and prediction customer charge change, improve prediction accuracy, Reasonable disposition for platform district distribution transformer capacity lays the foundation, and improves power distribution station economic operation level and construction efficiency.

Accompanying drawing explanation

The present invention is further described below in conjunction with the accompanying drawings.

Fig. 1: customer charge Forecasting Methodology application drawing in the embodiment of the present invention；

Fig. 2: genetic algorithm flow chart in the embodiment of the present invention；

Fig. 3: the non-sequential arbitrary sampling method flow chart in Monte Carlo in the embodiment of the present invention.

Detailed description of the invention

Embodiments of the invention are described below in detail, and the example of described embodiment is shown in the drawings, the most identical or Similar label represents same or similar element or has the element of same or like function.Describe below with reference to accompanying drawing Embodiment is exemplary, it is intended to is used for explaining the present invention, and is not considered as limiting the invention.

A kind of based on information about power the customer charge Forecasting Methodology that the present invention provides, it is possible to convenient, fast prediction user peak is born The load variations of lotus period, improves prediction accuracy, and the reasonable disposition for platform district distribution transformer capacity lays the foundation, and raising is joined Radio area economic operation level and construction efficiency.

One as it is shown in figure 1, the concrete steps of customer charge Forecasting Methodology include:

One) obtain the kind set DQ of user household electrical equipment, power set PDQ, magnitude-set NDQ, use at the beginning of probability Beginning representative value matrix PB⁰, the peak load period moment set T and the power consumption set Q of user's peak load period.

1, the kind set DQ={dq of household electrical appliance₁,dq₂,...,dq_j,...,dq_n, n is the type sum of household electrical appliance, dq_jFor The title of jth type household electrical appliance.

2, the power set PDQ={pdq of household electrical appliance₁,pdq₂,...,pdq_j,...,pdq_n, pdq_jFor jth type household electric The mean power of device.

3, the magnitude-set NDQ={ndq of household electrical appliance₁,ndq₂,...,ndq_j,...,ndq_n, ndq_jFor jth type household electrical appliance Quantity.

4, the moment set T={t of peak load period₁,t₂,...,t_i,...,t_m, t_mIt is the moment value of 24 times processed, 1≤t_m≤24。

5, the use probability initial representative value matrix of household electrical applianceFor t_iTime Carve jth type household electrical appliance dq_jUse probability representative value initial value.

6, the power consumption set Q={q of user's peak load period₁,q₂,...,q_k,...,q_d, q_kFor kth day user's peak hours The power consumption of section, d is total natural law that intelligent electric meter gathers described power consumption.

Two) user's moment sequence number i=1 of peak load period in a day, 1≤i≤m in the present embodiment are set.

Three) probability representative value matrix PB will be used^i-1Assignment is to using probability representative value matrix PBⁱIn.

PB^i-1For t_i-1The household electrical appliance in moment use probability representative value matrix, PBⁱFor t_iThe household electrical appliance in moment use probability typical case Value matrix.

Four) probability representative value matrix PB is used according to current household electrical applianceⁱ, carry out the non-sequential stochastic sampling in Monte Carlo, meter Calculate the load value p in each moment of this user peak load period_i, build load curve sequential value set Pⁱ={ p₁,p₂,...,p_i,...,p_m}。

Five) use probability representative value matrix PB is optimizedⁱ。

As in figure 2 it is shown, the present embodiment is used genetic algorithm optimization t_iThe use probability representative value matrix PB of moment household electrical applianceⁱ, tool Body step is:

1, setting the initial value of iterations h as 0, the maximum of iterations h is maxnum, and the quantity of population scale is ZQnum。

2, according to gene code strategy, by t_iThe use probability representative value matrix PB of moment household electrical applianceⁱIn each use probability to enter The chromosome coding that row is corresponding, produces initial population ZQ⁰。

3, iterations h=h+1 is set.

4, by population ZQ^h-1Data assignment to population ZQ^h。

ZQ^h-1It is the population of the h-1 time iteration generation, ZQ^hIt it is the population of the h time iteration generation.

5, population ZQ is set^hThe initial value of middle chromosome marker position l is 0.

6, chromosome marker position l=l+1 is put.

7, from population ZQ^hThe l chromosome of middle taking-up, carries out Gray code to this chromosome, obtains t_iMoment all types of household electric The use probability of device, such as, obtain t_iThe use probability of moment jth type household electrical appliance

8 as it is shown on figure 3, according to current t_iThe use probability of moment all types of household electrical appliance, sequential takes out at random with Monte Carlo is non- Quadrat method carries out stochastic sampling and obtains t_iThe household electrical appliance load value p of moment user_i。

9, the t obtained according to step 8_iThe household electrical appliance load value p of moment user_i, update current loads Curve Sequences value set Pⁱ={ p₁,p₂,...,p_i,...,p_m}。

10, according to current t_iThe load curve sequential value set P in momentⁱ={ p₁,p₂,...,p_i,...,p_m, calculate user in one day Load average value P av of peak load period household electrical appliance,

11, the average load set of user's peak load period is calculated according to power consumption set Q PG={pg₁,pg₂,...,pg_k,...,pg_d}。

Wherein, pg_kFor the average load value of kth day user, pg_k=q_k/ m, d are that intelligent electric meter gathers the total of user power utilization amount Natural law.

12, fitness function value ε of l chromosome is calculated,

If 13 chromosome marker position l ＜ ZQnum, then return step 6；If chromosome marker position l >=ZQnum, then perform step Rapid 14；

14, select.

Wherein, selection is fitness function value ε according to chromosome each in population, chooses fitness function value ε less ZQnum chromosome；By chromosome according to the ascending arrangement of the numerical value of fitness function value ε in the present embodiment, choose the 1st Step 15 is performed to the ZQnum chromosome；

15, to population ZQ^hCarry out parents' Shuangzi single-point gene Integral cross computing.

16, variation.

Wherein, variation is to control whether chromosome makes a variation according to aberration rate, when needing to make a variation, randomly chooses needs The gene of variation.After determining the gene needing variation, it should dependence and mutex relation situation according to gene make a variation.

17, according to the chromosome Population Regeneration ZQ after variation^h。

If 18 iterations h ＜ maxnum, then return step 3；If iterations h >=maxnum, then by fitness function The chromosome of value ε optimum carries out Gray code, obtains t_iThe use probability optimization value of moment household electrical appliance.

19, according to above-mentioned t_iThe use probability optimization value of moment household electrical appliance, determines the use probability allusion quotation of the household electrical appliance after optimization Offset matrix PBⁱ。

Six) moment sequence number i=i+1 of peak load period is set.

Seven) if i≤m, then step 3 is returned), probability representative value matrix PB will be used^i-1Assignment is to using probability representative value matrix PBⁱIn；If i is ＞ m, then according to the use probability representative value matrix PB after optimizingⁱThe all kinds household electrical appliance having with user Quantity calculate prediction customer charge value.

Two as it is shown on figure 3, step one in the present embodiment), five) and seven) the non-sequential arbitrary sampling method meter in middle Monte Carlo Calculate user at t_iThe load value p of moment household electrical appliance_iConcretely comprise the following steps:

1, the initial value of stochastic sampling number of times a is set as 0.

2, stochastic sampling number of times a=a+1 is set.

To household electrical appliance at t_iThe running status in moment carries out a time stochastic sampling, obtains jth type household electrical appliance dq_jAt t_i The a time stochastic sampling running status coefficient in moment

If the random number of a time stochastic samplingThenHousehold electrical appliance off-duty；If a time stochastic sampling Random numberThenHousehold electrical appliance run；For t_iThe use probability of moment household electrical appliance.

3, user is calculated at t_iThe load value of a time stochastic sampling in moment

If 4 stochastic sampling number of times a ＜ RS, then return step 2；If stochastic sampling number of times a >=RS, then calculate RS time at random The household electrical appliance load value p of the lower user of sampling_i；RS is largest sample number of times.

Three, the load forecasting method using the present invention to provide, carries out the concrete mistake of load prediction peak load in the summer period to user Cheng Wei:

One) kind set DQ, the power set P of household electrical appliance are determined_DQ, magnitude-set N_DQ, use the initial representative value of probability Matrix, the moment of user's peak load period set T and user power utilization duration set Q.

1, the kind set of household electrical appliance:

DQ={dq₁,dq₂,dq₃,dq₄,dq₅,dq₆,dq₇,dq₈,dq₉,dq₁₀,dq₁₁,dq₁₂,dq₁₃,dq₁₄,dq₁₅,dq₁₆}:

dq₁Represent lamp lighting apparatus, dq₂Represent electromagnetic oven, dq₃Represent electric cooker, dq₄Represent electric kettle, dq₅Represent micro- Ripple stove, dq₆Represent smoke exhaust ventilator, dq₇Represent high-grade electric cooking appliance (including the high end equipments such as electric baking pan), dq₈Represent electric fan, dq₉ Represent air-conditioning, dq₁₀Represent space heater, dq₁₁Represent television set, dq₁₂Represent electric refrigerator, dq₁₃Represent washing machine, dq₁₄Table Show electric heater, dq₁₅Represent home computer, dq₁₆Represent electric bicycle.

2, the power set PDQ={pdq of household electrical appliance₁,pdq₂,...,pdq₁₆}。

3, the magnitude-set NDQ={ndq of household electrical appliance₁,ndq₂,...,ndq₁₆}。

4, the moment set T={9,10,11,17,18,19,20,21} of peak load period.

5, the use probability initial representative value matrix of household electrical appliance

6, user power utilization duration set Q={q₁,q₂,...,q₆₂, the present embodiment is chosen user 7 and August that intelligent electric meter gathers Power consumption.

Two) user moment sequence number i=1 of peak load period, 1≤i≤8 in the present embodiment in one day are set.

Three) probability representative value matrix PB will be used^i-1Assignment is to using probability representative value matrix PBⁱIn.

Four) probability representative value matrix PB is used according to current household electrical applianceⁱ, carry out the non-sequential stochastic sampling in Monte Carlo, meter Calculate the load value p in each moment of this user peak load period_i, form current load curve sequential value set Pⁱ={ p₁,p₂,...,p_i,...,p_m}。

Five) according to this user's July in summer and the power consumption Q={q of the peak load period of every day in August₁,q₂,...,q₆₂And it is current negative Lotus Curve Sequences value set Pⁱ={ p₁,p₂,...,p_i,...,p_m, use genetic algorithm progressively to t_iMoment point uses probability representative value square Battle array PBⁱIt is optimized.

Six) update current household electrical appliance according to the use probit after optimizing and use probability representative value matrix PBⁱ,

Seven) moment sequence number i=i+1 of peak load period is put.

Eight) judge moment sequence number i of peak load period whether more than 8, if moment sequence number i of peak load period less than or Equal to 8, go to step three)；If i is more than 8, go to step nine).

Nine) use, according to the household electrical appliance after optimizing, all kinds household electrical appliance that probability representative value matrix and this user have Quantity, according to the non-sequential stochastic sampling in Monte Carlo, household electrical appliance all to this user at the running status stochastic sampling in moment, Predict the load value of this user peak period in summer.

Finally should be noted that: described embodiment is only some embodiments of the present application rather than whole embodiments. Based on the embodiment in the application, those of ordinary skill in the art obtained under not making creative work premise all its His embodiment, broadly falls into the scope of the application protection.

Claims (5)

1. a customer charge Forecasting Methodology based on information about power, it is characterised in that described method includes:

Step 1: obtain the kind set DQ of user household electrical equipment, power set PDQ, magnitude-set NDQ, use probability Initial representative value matrix PB⁰, the peak load period moment set T and user peak load period power consumption set Q；

Step 2: setting user's moment sequence number i=1 in the peak load period, 1≤i≤m, m are the moment of peak load period Sum；

Step 3: probability representative value matrix PB will be used^i-1Assignment is to using probability representative value matrix PBⁱIn；Described PB^i-1For t_i-1 The household electrical appliance in moment use probability representative value matrix, PBⁱFor t_iThe household electrical appliance in moment use probability representative value matrix；

Step 4: according to described use probability representative value matrix PBⁱCalculate the load value in each moment, structure in user's peak load period Build load curve sequential value set Pⁱ={ p₁,p₂,...,p_i,...,p_m, p_iFor t_iThe load value of moment household electrical appliance；

Step 5: with using probability representative value matrix PB described in genetic algorithm optimizationⁱ；

Step 6: moment sequence number i=i+1 of peak load period is set；

Step 7: if i≤m, then return step 3；If i is ＞ m, then according to the use probability representative value of the household electrical appliance after optimizing Matrix PBⁱWith the quantity of all kinds household electrical appliance that user has, the non-sequential arbitrary sampling method in Monte Carlo is used to calculate pre- Survey the load value of user.

2. the method for claim 1, it is characterised in that the kind set of household electrical appliance described in described step 1 DQ={dq₁,dq₂,...,dq_j,...,dq_n, n is the type sum of household electrical appliance, dq_jTitle for jth type household electrical appliance； The power set PDQ={pdq of described household electrical appliance₁,pdq₂,...,pdq_j,...,pdq_n, pdq_jFor jth type household electrical appliance Mean power；The magnitude-set NDQ={ndq of described household electrical appliance₁,ndq₂,...,ndq_j,...,ndq_n, ndq_jFor jth type The quantity of household electrical appliance；The moment set T={t of described peak load period₁,t₂,...,t_i,...,t_m, t_mIt it is the moment of 24 times processed Value, 1≤t_m≤24；

The use probability initial representative value matrix of described household electrical appliance

For t_iMoment jth type household electrical appliance dq_jUse probability representative value initial value.

3. the method for claim 1, it is characterised in that user power utilization duration set in described step 1 Q={q₁,q₂,...,q_k,...,q_d, q_kFor the power consumption of kth day user, d is total natural law that intelligent electric meter gathers described power consumption；

Described step 4 calculates t with the non-sequential arbitrary sampling method in Monte Carlo_iThe load value p of moment household electrical appliance_i。

4. the method for claim 1, it is characterised in that use genetic algorithm optimization t in described step 5_iMoment household electric The use probability representative value matrix PB of deviceⁱ, including:

Step 5-1: setting the initial value of iterations h as 0, the maximum of iterations h is maxnum, the quantity of population scale For ZQnum；By t_iThe use probability representative value matrix PB of moment household electrical applianceⁱIn each chromosome using probability to carry out correspondence Coding, produces initial population ZQ⁰；

Step 5-2: described iterations h=h+1 is set；

Step 5-3: by population ZQ^h-1Data assignment to population ZQ^h；ZQ^h-1It is the population of the h-1 time iteration generation, ZQ^h It it is the population of the h time iteration generation；Set population ZQ^hThe initial value of middle chromosome marker position l is 0；

Step 5-4: described chromosome marker position l=l+1 is set；

Step 5-5: the l chromosome is carried out Gray code, obtains t_iThe use probability of moment all types of household electrical appliance；

Step 5-6: according to described t_iThe use probability of moment all types of household electrical appliance, with the non-sequential arbitrary sampling method in Monte Carlo Obtain t_iThe load value p of moment household electrical appliance_i, and update described load curve sequential value set Pⁱ={ p₁,p₂,...,p_i,...,p_m}；

Step 5-7: calculate user's bearing at peak load period household electrical appliance according to the load curve arrangement set P after described renewal Lotus average value P av, $\mathrm{Pav}=\left(\underset{i=1}{\overset{m}{\mathrm{\Σ}}}{p}_i\right)/m;$

Step 5-8: obtain the average load set of user's peak load period according to power consumption set Q PG={pg₁,pg₂,...,pg_k,...,pg_d}；pg_kFor the average load value of kth day user, d is that intelligent electric meter gathers user power utilization Total natural law of amount；

Step 5-9: calculate fitness function value ε of l chromosome,

Step 5-10: if chromosome marker position l ＜ ZQnum, then return step 5-4；If chromosome marker position l >=ZQnum, Then perform step 5-11；

Step 5-11: by described chromosome according to the ascending arrangement of the numerical value of fitness function value ε, choose the 1st to ZQnum Step 5-12 is performed after individual chromosome；

Step 5-12: to population ZQ^hCarry out parents' Shuangzi single-point gene Integral cross computing；

Step 5-13: chromosome is made a variation, and update described population ZQ according to the chromosome after variation^h；

Step 5-14: if iterations h ＜ maxnum, then return step 5-2；If iterations h >=maxnum, then by institute The chromosome stating fitness function value ε optimum carries out Gray code, obtains t_iThe use probability optimization value of moment household electrical appliance.

5. the method as described in claim 1,3 or 4, it is characterised in that in step 7, the non-sequence in described Monte Carlo Pass through arbitrary sampling method and calculate t_iThe load value p of moment household electrical appliance_i, including:

Step 7-1: set the initial value of stochastic sampling number of times a as 0；

Step 7-2: stochastic sampling number of times a=a+1 is set；

To household electrical appliance at t_iThe running status in moment carries out a time stochastic sampling, obtains jth type household electrical appliance dq_jAt t_i The a time stochastic sampling running status coefficient in moment

If the random number of a time stochastic samplingThenIf the random number of a time stochastic sampling Then For t_iThe use probability of moment household electrical appliance；

Step 7-3: calculate user at t_iThe load value of a time stochastic sampling in moment

Step 7-4: if stochastic sampling number of times a ＜ RS, then return step 1-2；If stochastic sampling number of times a >=RS, then calculate RS Described load value p under secondary stochastic sampling_i；

DescribedRS is largest sample number of times.